摘要
基于计算核和图形接口能够支持GPU通用计算的特点,给出了一个规范的四层编程模型。本文主要研究了在系统抽象层中对计算核和图形接口层的图形操作抽象成通用计算的过程,其目的是屏蔽图形操作的痕迹,使在应用层执行屏蔽图形绘制程序,图形接口能够被抽象成GPU进行通用计算操作,提高面向GPGPU绘制语言的可描述性和规范化,并通过检测模块的设置检测运行环境,以提高效率。
A four layer programming model is given in the paper according to the characteristics of computing core and graphic interface to support GPU general computation.Graphic operating of graphic interface layer and computing core in system abstraction layer is abstracted to the general computing process,to mask the trace of graphic operation and shield the program of graph drawing in application layer.Its purpose transforms graphic interface to GPU general computing to improve specification and normalization of graphic language based on-GPGPU.Furthermore,it can detect the operating surrounding to improve efficiency by setting detecting model.
出处
《河南科技大学学报（自然科学版）》
CAS
北大核心
2009年第3期51-54,共4页
Journal of Henan University of Science And Technology:Natural Science
基金
国家自然科学基金项目(60873030)
关键词
图形处理器
通用计算
绘制语言
模型
Graphics processing unit
General purpose computation
Shading language
Model
